Amazon Stock Price Forecast: AMZN Hits $253 on Trainium Ramp, $100B Anthropic Deal — $293 Price Target in Focus

Amazon.com (NASDAQ:AMZN) is quietly executing one of the most complete business transformations in mega-cap equity history, and the tape at $253.44 with a 1.41% session gain reflects only a fraction of what's actually happening inside the corporate structure. The stock has ripped roughly 8% year-to-date through Wednesday's print, making AMZN the second-best-performing Magnificent Seven name behind only Broadcom (AVGO) and decisively outpacing the S&P 500's roughly 2% gain — a complete reversal from the positioning heading into Q4 2025, when Amazon was the worst-performing Mag7 name burdened by tariff exposure, de minimis vulnerabilities on its China-dependent retail sourcing, and the shock of its $200 billion capex guide that shattered sentiment and produced an 11% post-print drawdown. What's changed since then isn't the fundamental business — it's the market's recognition of what Amazon has built beneath the capex headline. The AWS hyperscale engine is running at a $142 billion annualized revenue run rate with 24% year-over-year growth. Trainium custom AI silicon plus Graviton generalized compute combined are producing over $20 billion in annualized revenue with triple-digit percentage growth. The Anthropic partnership has locked in $100 billion in committed AWS spending over the next decade. More than one million robots are operating in fulfillment centers projected to save $10 billion annually when scaled. Project Leo (formerly Kuiper) is launching satellite broadband backed by the $11.6 billion Globalstar acquisition and an Apple partnership. Q1 earnings land April 29 after the close, and the setup into that print combined with the structural catalysts sitting beneath the surface produces one of the cleanest risk-adjusted mega-cap setups in the current market. The positioning verdict matters and deserves careful reading — let's unpack every dimension driving the thesis.

The AWS Engine — $142 Billion Run Rate, 24% Growth, and the Capacity-Constrained Demand Signal

Amazon Web Services is doing something that deserves specific attention because the conventional narrative significantly understates what's actually happening. AWS generated $128.7 billion in FY2025 revenue versus Microsoft Azure's $75 billion and Google Cloud Platform's $58.7 billion, giving Amazon hyperscale cloud revenue that essentially matches its two largest competitors combined. The current annualized run rate has climbed past $142 billion growing at 24% year-over-year, and CEO Andy Jassy's commentary in the 2026 shareholder letter flagged something the market hasn't fully internalized — AWS has capacity constraints that are producing genuine unserved demand. Jassy disclosed that two large AWS customers have requested to buy the entirety of Amazon's Graviton instance capacity for 2026, requests the company is declining because of obligations to other clients. That's not a symptom of a business losing share; it's a symptom of a business that cannot supply enough compute to meet existing demand. The competing narrative that Google Cloud is "gaining share" growing from 13% to 14% and Azure moving from 20% to 21% while AWS contracts to 28% market share is technically accurate but materially misleading — AWS is operating at a scale where absolute revenue growth dwarfs percentage market share movements, and the 24% growth rate on a $128.7 billion base adds more absolute dollars to the top line than either competitor adds in total. Jassy has explicitly telegraphed that AWS "could be growing faster" if capacity constraints were resolved, which means the 24% rate represents a supply-constrained floor rather than a demand-constrained ceiling.

Anthropic's $100 Billion Commitment and Project Rainier's 500,000 Trainium Deployment

The Anthropic relationship is the single most important commercial development in the AWS ecosystem over the past eighteen months, and the structure of the deal architecture is worth understanding precisely. Amazon made an initial $8 billion investment in Anthropic during the AI startup's early cycle, then expanded the commitment on April 20, 2026 with an additional $5 billion plus up to $20 billion more in future tranches — bringing potential total exposure to $33 billion. In return, Anthropic committed to spending $100 billion over the next decade on AWS technologies and specifically reserved its backlog of Amazon's custom Trainium chips as the dedicated compute substrate for Claude model training and deployment. The partnership has secured up to 5 gigawatts of dedicated compute capacity. Anthropic's own annualized revenue run rate grew from $9 billion to more than $30 billion by year-end 2025, outpacing OpenAI's $25 billion ARR and flipping the competitive landscape in AI model monetization. Claude's daily average user rate "more than doubled" between February and March 2026, and BNP Paribas analysts led by Nick Jones documented ChatGPT losing both web traffic and mobile app share to Claude in March. The commercial implications transmit directly into AWS revenue: Project Rainier, the $11 billion data center campus in Indiana, deploys 500,000 Trainium2 chips exclusively for Anthropic's Claude training and inference workloads. Wells Fargo raised AWS's 2026 growth outlook from +18-19% to +22% specifically on the back of Project Rainier, which is expected to ramp to full 2.2 gigawatt capacity by year-end and add an estimated $14 billion in annual AWS revenue when operating at scale.

Custom Silicon — Trainium and Graviton at $20 Billion Run Rate With Triple-Digit Growth

The custom chip division is quietly generating the structural cost advantage that will define AWS margins over the next five years, and the numbers are significantly larger than most analysts recognize. Graviton, Trainium, and Nitro collectively produce more than $20 billion in annualized revenue growing at triple-digit percentage rates year-over-year. Oakoff Investments' tracking places the combined ARR at over $10 billion for Trainium and Graviton alone before accounting for third-party sales potential. Jassy's specific framing on the margin impact matters: custom silicon at scale will save Amazon "tens of billions of capex dollars per year" and provide "several hundred basis points of operating margin advantage versus relying on others' chips for inference." That's a structural EBIT margin expansion embedded directly in the business model that compounds year over year. Trainium2 delivers 30%-40% better price-performance than equivalent GPU-based EC2 P5e and P5en instances or Nvidia H100 references, with more than 100,000 companies using it as the majority underpinning of Bedrock. Trainium3, which began shipping in early 2026, delivers up to 40% additional price-performance improvement over Trainium2 and has nearly all of its supply committed by mid-2026. Trainium4 is already attracting strong interest despite not yet being built. Jassy has publicly suggested that if the chip division operated as a standalone company, it would easily achieve a $50 billion run rate on its own — which indicates Amazon could eventually sell hardware directly to third parties and compete head-on with Nvidia in the accelerator market. The strategic implication: Amazon has built a semiconductor business inside AWS that individually rivals most mid-cap silicon names, and the market is pricing essentially none of that optionality into the stock.

The OpenAI Pivot — $50 Billion Deal and 2 Gigawatts of Trainium Capacity

One of the more surprising developments of early 2026 was OpenAI's pivot toward AWS infrastructure despite its historical Microsoft Azure alignment. Amazon announced a $50 billion investment deal with OpenAI that made Amazon the exclusive provider of OpenAI's new AI agent builder, Frontier. The deal structure includes Amazon supplying OpenAI with 2 gigawatts of Trainium capacity, which expands the chip's deployment footprint beyond Anthropic and validates that competing AI labs view the price-performance advantage as sufficient to justify infrastructure migration. That's significant because it demonstrates Trainium's competitive position isn't dependent on a single anchor customer — the technology is proving viable across multiple large-scale AI training workloads. Between Anthropic's Project Rainier deployment at 500,000+ Trainium2 chips and OpenAI's 2 GW Trainium allocation, Amazon has secured the two most important AI model builders in the industry on its custom silicon, a structural position that neither Google nor Microsoft has matched.

The Robotics Flywheel — 1 Million Robots and $10 Billion Annual Savings

The physical fulfillment automation story sitting inside Amazon deserves specific focus because it's creating margin expansion that the market consistently underprices. Amazon now operates more than one million robots in its fulfillment centers handling stowing, picking, sorting, and intra-facility transport. Morgan Stanley's analysis identifies that fulfillment costs make up roughly 20% of retail revenue, and scaled robotics implementation could save more than $10 billion annually when fully deployed — transforming fulfillment from a major operational expense into a genuine margin contributor. The Q4 2025 trend showed unit growth outpacing shipping cost growth by approximately 200 basis points, which captures the early-stage margin benefit already materializing. Beyond warehouse robotics, Amazon is preparing to launch Zoox as a commercial robotaxi service in Austin and Miami later in 2026, with the network eventually targeting the 390,000 drivers and independent delivery partners who currently handle last-mile delivery. The strategic integration matters: Amazon's AWS and Bedrock AI capabilities feed into the robotics training pipeline, which automates warehouse operations and eventually last-mile delivery, which compresses the cost structure on every retail item sold, which generates the capital that funds continued AI and robotics investment. This is the classic flywheel structure that produces compounding advantage over time, and it's reaching the inflection point where investors will see it in operating margin expansion rather than just narrative.

Project Leo and the $11.6 Billion Globalstar Acquisition — The Connectivity Layer

Amazon's orbital network expansion through Project Leo (rebranded from Project Kuiper) represents the fourth vertical in the integrated ecosystem, and the strategic architecture matters even though the near-term execution risk is real. The $11.6 billion Globalstar acquisition provides Amazon with the spectrum necessary to accelerate direct-to-device capabilities for Project Leo, enabling cellular network access in any region. A new partnership with Apple will see Amazon Leo power satellite services for iPhone and Apple Watch once launched, which instantly places Amazon's infrastructure in the pockets of hundreds of millions of users. The connectivity layer creates vertical integration that no competitor can match — Amazon owns the retail storefront, the cloud infrastructure, the custom silicon, the robotics automation, and now the communications network. Execution risk is meaningful: the FCC requires Amazon to have half of its 3,232-satellite constellation in orbit by July 2026, and as of March 2026 only 212 satellites were deployed. That's significantly behind schedule and could result in loss of critical spectrum licenses if launch milestones are missed. Management flagged that Q1 would be impacted by roughly $1 billion in costs tied to the Leo buildout, which will be watched closely on the April 29 earnings call as investors look for clarity on the OPEX-to-capitalized-asset shift as the network nears commercial launch later this year.

The Advertising Business — $85 Billion Run Rate Growing at 23% Year-Over-Year

The advertising business is the quietly compounding profit engine that deserves more attention than it typically receives in Amazon analysis. The ads division is growing at 23% year-over-year with an $85 billion annualized revenue run rate as of Q4 2025, and Amazon recently integrated ads into Prime Video which creates an incremental high-margin revenue stream that flows essentially entirely to EPS because it requires minimal additional cost. Advertising is structurally the highest-margin segment inside Amazon's business mix — gross margins well north of 70% translate into outsized operating leverage as the division scales. For Q1 2026 reporting, the comparison base from Q1 2025 sets up favorably because 2025 started from a relatively modest baseline before the Prime Video ad integration scaled, meaning upside surprises in the advertising line should be possible even without heroic assumptions on underlying ad market dynamics. The embedded flywheel benefit is that ad revenue funds further cloud and silicon investment, which enhances the retail platform, which attracts more consumer engagement, which expands the addressable advertising inventory — exactly the kind of self-reinforcing structure that produces compounding value creation.